Any fixes for this: https://github.com/racket/typed-racket/issues/455 ?

what should the type of create-generator be?

I’m not sure about racket/generator internals. Is there a theoretical limitation in design of the type system that generators cannot be handled?

that all depends on what create-generator does

I haven’t looked at it

I had a quick look, it’s a bit dense and uses continuations for the implementation https://github.com/racket/racket/blob/5147771b04ed1d6d825c049c0870bc6d7c7de15b/racket/collects/racket/generator.rkt#L87

yes, it’s pretty tricky code

Anyways, right now I’m leaving untyped the part of my code which needs to use generators, but this is just another “it would be nice to have another base library in typed racket” request.

So YIL (yesterday I learned) about this protocol where macros can emit info messages to an online-check-syntax logger, and the data member has a value to describe a tooltip: https://github.com/greghendershott/racket-mode/issues/451

I can’t seem to find documentation for this, so I’m learning about it “forensically”.

So far, what I “know” is that there are two flavors of the data: https://github.com/greghendershott/racket-mode/commit/2c55a4bf1252942d8450cedd10788730b4e43876#diff-f28781069d424b4bc3c2c4261f0f34abR51-R77

https://docs.racket-lang.org/tools/Cooperating_with_Background_Check_Syntax.html?q=check-syntax

@samth Thanks. I had overlooked that. But. That just says there is a list of syntax objects. For each object, there is more to the story. Apparently there’s a syntax property mouse-over-tooltips. Its value comes in at least two flavors. One is a simple (vector beg end text). That’s used by e.g. Pie. Another thing I’ve seen is Typed Racket supplies a list of vectors. Furthermore the text isn’t a string, it’s a thunk to call to get the string.

If that’s the whole story, fine, I figured it out. Maybe I could even submit a PR to extend the docs you pointed me to.

I’m just wondering if (a) I really have it correct and (b) are there some 3rd or more “flavors” of the value I haven’t yet discovered. :slightly_smiling_face:

those docs are here https://docs.racket-lang.org/tools/Check_Syntax.html?q=check-syntax#%28part._.Syntax_.Properties_that_.Check_.Syntax_.Looks_.For%29

i think we should fix the lack of indexing of those symbols

Ugh I’ve seen that page but somehow managed to overlook the tooltips part at the end. Thanks!

Indexing would help, as would maybe a link from that first page you showed me, to the second.

When I finish this up, I’ll see if I could supply a PR to do those things for the docs….

Thanks again and big :thumbsup: that this protocol was actually documented.

if you look at scribblings/reference/mz.rkt, the indexed-racket function is what you want to use

I don’t know if that’s available in some more sensible location

https://pkg-build.racket-lang.org/server/built/fail/polyglot.txt

I’m guessing this really is temporary, but I’m putting this here in case there’s a network configuration or firewall issue I don’t know about.

pkg: Contacting <http://github.com\|github.com> pkg: Network error; retrying after 0.1s pkg: Contacting <http://github.com\|github.com> pkg: Network error; retrying after 0.2s pkg: Contacting <http://github.com\|github.com> pkg: Network error; retrying after 0.4s pkg: Contacting <http://github.com\|github.com> pkg: Network error; retrying after 0.8s pkg: Contacting <http://github.com\|github.com> pkg: Network error; retrying after 1.6s pkg: Contacting <http://github.com\|github.com> tcp-connect: connection failed; host not found address: <http://github.com\|github.com> port number: 443 system error: Temporary failure in name resolution; gai_err=-3 &lt;snip&gt;

> package builds are performed on a VM without network connectivity. Never mind.

Is this for testing or for documentation?

If it’s documentation, you can use make-log-based-eval to workaround the issue

But if it’s for testing, then it won’t work. You can test however if you are in the pkg-build environment via PLT_PKG_BUILD_SERVICE env var, and bail out if that’s the case.

Metapackage, so it has an implies for doc, test, and lib packages. I tried to let each of them pin a GitHub URL dependency to give myself room to publish a breaking change to another package. It’s on me for missing the note about network connectivity.

It’s not on you. The error message does not explain the problem.